A new theropod dinosaur from the early cretaceous (Barremian) of Cabo Espichel, Portugal: Implications for spinosaurid evolution

Spinosaurids are some of the most enigmatic Mesozoic theropod dinosaurs due to their unique adaptations to aquatic environments and their relative scarcity. Their taxonomy has proven to be especially problematic. Recent discoveries from Western Europe in general, specifically Iberia, provide some of the best specimens for the understanding of their phylogeny, leading to the description of the spinosaurid Vallibonavenatrix cani and the recognition of the Iberian dinosaur Camarillasaurus cirugedae as one of them. Portuguese associated spinosaurid remains (ML1190) from the Papo Seco Formation (early Barremian) were previously assigned to Baryonyx walkeri but new material recovered in 2020 along with new phylogenetic analyses suggests a different phylogenetic placement, making their revision necessary. Here we show that these remains are not attributable to Baryonyx walkeri, but to a new genus and species, Iberospinus natarioi, gen. et sp. nov. The new taxon is characterized by the presence of a single Meckelian foramen in the Meckelian sulcus, a straight profile of the ventral surface of the dentary and a distal thickening of the acromion process of the pubis between other characters. Iberospinus natarioi is recovered as a sister taxon of the clade formed by Baryonyx and Suchomimus, and outside Spinosaurinae when Vallibonaventrix cani is excluded from the analysis. The description of this taxon reinforces Iberia as a hotspot for spinosaur biodiversity, with several endemic taxa for the region. As expected for the clade, the dentary displays a highly vascularized neurovascular network. The morphometric analysis of parts of the skeleton (pedal phalanx and caudal vertebrae, among others) shows an intermediate condition between basal tetanurans and spinosaurines.

The vertebra is roughly as tall as long. The anterior facet is concave and slightly oval, with an anterior cotile of about 7 mm of diameter in the middle.The anterior end of the neural canal extends anteriorly over the concave surface of the centrum. The area where the neural arch was attached to the centrum is extremely rough. This rugose area is wider mediolaterally in the anterior portion of the neural canal. The neural canal is hourglass shaped, wider mediolaterally at both anterior and posterior ends (although is slightly wider in the anterior end). It has some ridges running anterodistally across its ventral surface. Ventrally the vertebra presents a clearly marked groove in the midline, whose edges end posteriorly in the chevron facets. The grove is more apparent in the posterior part of the vertebra; meanwhile it almost disappears in the anterior part. There is a small protuberance or bulge in the medial part of the groove, on its posteriormost part.
The posterior facet is mostly flat (although it is the most damaged part of the vertebra) without any signs of cotile and is about as oval as the anterior facet.

ML1190-17
This vertebra is one of the most damaged ones of the vertebral series. The neural arch is completely missing, the right anterior part of the centrum totally disappeared and the rims of both central facets are highly eroded (Fig S2). Nevertheless, it is still possible to see the posterior part of the neural canal, whose edges tapper medially towards the middle of the vertebra, as well as ventrally as the canal goes deeper into the centrum. The ventral half of the posterior facet is straight, meanwhile the dorsal half is slightly concave. It possesses a small foramen of about 2,5 mm in diameter in the dorsal half of the left side of the centrum, more or less in the middle between anterior and posterior facet rims.There is a clear groove running ventrally across the midline of the centrum, which almost disappears in the anterior part. Both edges that limit it dive more pronouncedly towards the groove in the posterior part.

ML1190-16
This vertebra was already described by (1) although briefly, and was interpreted as midanterior caudal. It possesses the neural arch, although it lacks the transverse processes, the neural spine, both prezygapophyses and postzygapophyses and only the hyposphene is still mostly intact. Also, the right posterior ventral part of the centrum and the ventral part of the anterior facet are eroded and lightly damaged. The whole vertebra is slightly crushed mediolaterally and the left side is better preserved, a situation similar to that of the entire vertebral series (Fig S3). n.c, neural canal, v.dep, ventral depression in the neural arch, dia, diaphysis, n.s, broken neural spine, hyp, hypantrum, spnpr.fo, spinoprezygapophyseal fossa, v.gr, ventral groove, che.fa, chevron facets.
The anterior facet of the centrum is clearly concave and oval, without any distinguishable cotile on it. The ventral margin of the neural canal in this anterior facet is mostly rectilinear mediolaterally.
The neurocentral suture is clearly visible in the anterior half of the vertebra, but it becomes less marked as it goes backwards, being almost invisible near the posterior rim of the facet.
This suture also tapers ventrally as it runs along the centrum, in a similar manner to ML1190-15. Finally, there is a shallow depression in the middle of the lateral surface of the centrum, but not any signs of pleurocoels.
In ventral view the vertebra is clearly hourglass shaped and, with the ventral groove clearly

ML1190-22
This vertebra has not previously been described. It is strongly crushed mediolaterally, with the right ventral part of the centrum damaged. It still preserves the neural arch, although with just the eroded and damaged right prezygapophysis, all the other characters except the base of the transverse processes being lost or eroded ( Fig S4). The ventral groove runs across the midline of the vertebra, clearly marked, even in the anterior half, although the deformation of the whole vertebra might make it more pronounced. The right chevron facet is slightly depressed, although this could be distorted by the preservation. The preserved part of the posterior central facet is slightly concave.
In the neural arch, the right prezygapophysis elevates in an angle of about 70º from the dorsal margin of the centrum. The spinoprezygapophyseal laminae are not well preserved but the spinoprezygapophyseal fossa delimited by them is much shallower and less elongated anteroposteriorly than in ML1190-16. In the anteroventral part of the base of the transverse processes there is a shallow depression, of bigger size than the small depression situated in the posteroventral part of them (albeit this last one is shallower than in ML1190-16).

ML1190-18
This vertebra was figured and interpreted as a mid posterior caudal vertebra by (1). The centrum is eroded in the left anteroventral part and in the left posterior part. In its left side, the centrum bears an oval mark with the major axis pointing anterodorsally-posteroventrally; measuring 21 mm in length and 11mm in width. This mark was already noted by (1) and attributed to the perimortem damage provoked by a predator or scavenger. The neural arch of the vertebra is still in place and the bases of both transverse processes and the neural spine are in place, in spite of this last feature being slightly deformed. The pre-and postzygapophyses are completely gone ( Fig S5). The anterior facet of the centrum is relatively square shaped, entirely concave. The ventral surface of the anteriormost part of the neural canal is slightly bent ventrally on its medial part.
The lateral surface of the centrum is relatively smooth, with some minor rugosities in the dorsal half. In the ventral view the vertebra is hour-glass shaped and its ventral surface presents a clear groove in the midline, although that is less marked than in previously described vertebrae.
The posterior facet of the centrum is even more square shaped than the anterior face, with the dorsal right part making an angle of almost 90º between the dorsal and right lateral rims.
The facet is concave, but only clearly on its dorsal half, with the ventral part more flat.
In the anterior dorsal part of the neural arch it is possible to see a fragment of the right spinoprezygapophyseal lamina, as well as a portion of the floor of the spinoprezygapophyseal fossa. There are two ridges running anteriorly from the anterior part of the base of the transverse processes. The ventral one finishes in the lateral wall of the neural canal opening, the dorsal one goes towards the spinoprezygapophyseal lamina, both ridges delimiting a fossa anterior to the transverse processes. The posterior part of the neural spine is visible, but not much information can be taken from it.

ML1190-21
This vertebra is unpublished. It is the most damaged one that could be recognized in the entire series. Only the postero-ventral left portion of it has part of the bone surface intact ( Fig   S6). The part that remains of the lateral surface of the vertebra has the characteristic rugosities towards the posterior part.
In the ventral surface the groove that runs across the midline is still distinguishable because there is part of the right ridge that delimited it still in place.
The posterior facet of the centrum is mainly completely eroded away, but the general pattern of the other vertebras can be recognized, with a ventral part more flattened meanwhile the part that remains of the dorsal half has a shape congruent with a more concave surface.

ML1190-20
This vertebra was already mentioned by (1). It comprises just the posterior half of a centrum, obliquely broken towards the posterior right part and missing the neural arch completely. The posterior right part is also eroded, including the small lateral surface visible (Fig S7). n.c, neural canal, v.gr, ventral groove, che.fa, chevron facets, geo, geode structure, k.ca, keel in the dorsal surface of the vertebral cavity.
The broken surface in the centrum shows the hollow interior of the vertebra, filled with a geode structure already noted by (1). This cross section shows a morphology that reflects that of the surface of the centrum, with a keel pointing ventrally from the dorsal surface of the cavity, in conformity with the depth that the neural canal reaches into the centrum.

ML1190-26
This unpublished vertebra is one of the best preserved of the specimen ML1190. The centrum is well preserved except for the anterior and posterior right lateral rims, which are slightly eroded. The neural arch is still in place, but is better preserved in the anterior part, with both prezygapophyses still in place as well as the base of neural spine and the bases of both transverse processes. The posterior part is worst preserved with the roof of the neural canal eroded as well as both postzygapophyses gone (Fig S8).

ML1190-23
This vertebra is unpublished and the neural arch is badly damaged. The centrum itself is well preserved (Fig S9). n.c, neural canal, v.ke, ventral keel, che.fa, chevron facets, pl.d, pleurocelic depression, pit.ri, pits and ridges in the floor of the neural canal.
The exposed floor of the neural canal presents a series of ridges and pits. The anterior facet of the centrum is concave; with the lateral rims slightly flexing outward, although the overall shape of the facet is square. In the dorsal half of the lateral surface of the centrum there is a shallow fossa across much of the lateral surface of the centrum (pleurocelic depression).
The ventral surface is mostly smooth although there is an extremely thin keel of about 25 mm long situated slightly posteriorly in the surface. The posterior facet has clearly 90º corners between the upper and lateral rims of the posterior surface, being the overall shape rectangular, taller than wider. The chevron facets seem to be asymmetrical, probably due to a certain degree of deformation or erosion, being the right one larger and more anteroposteriorly oriented.

ML1190-19
This vertebra was already mentioned in (1). It is the best preserved vertebra of the entire series. The centrum is completely preserved, with almost no erosion in the rims of the facets and the original surface mostly intact. The neural arch is more damaged, with cracks over most of its surface. The left prezygapophysis anterior part is eroded; there is only a hint of the anterior process of the neural spine and its posterior part, as well as the postzygapophysis are damaged (Fig S10). spinoprezygapophyseal laminae or fossa because of the conservation, but there is a small protuberance slightly situated in the anterior part of the vertebra, 6 mm tall and 5 mm wide herein interpreted as the cranial process of the vertebra also mentioned in a spinosaur from thailand (2). There are no transverse processes in this vertebra, but there are two smooth ridges only 2 or 3 mm wide, slightly anterior to the anteroposterior midpoint of the vertebra that could be homologous to the transverse processes preserved in more anterior vertebrae.
Therefore the smooth ridge that runs from their posterior part towards the postzygapophyses would be homologous to the diapostzygapophyseal lamina. The neural spine is about 14 mm wide in the base and is broken ~40 mm above the centrum. There is a triangular canal exposed by this breakage running posteroventrally across it, 9 mm tall and 6 mm wide in the ventral surface. Only the anteriormost part of the postzygapophyses is conserved. They overhang the posterior surface of the centrum in dorsal view and are separated ventrally for about 5 mm at their base.

ML1190-27
This vertebra is unpublished. It is relatively well preserved, although the right lateral rim of the anterior face of the centrum is heavily eroded. The neural arch is still in place, although the pre-and post-zygapophyses (as well as the neural spine) are totally eroded away ( Fig   S11). The anterior face of the centrum is concave and the lateral rim that remains is clearly bowed toward the lateral side. The lateral surface has a shallow pleurocentral depression, crossed by some longitudinal ridges. The ventral surface has a smooth and small keel across the midline; there are also some ridges and a small foramen in the anterior left ventral part. The posterior face of the centrum is also concave, the upper rim of it is mostly straight, however, the lateral rims converge slightly towards the ventral part and are slightly bowed, making this face taller than wider. The chevron facets are 13 cm high dorsoventraly.
The neural arch has lost all of its major features, but it is still possible to say that the vertebra had no transverse processes at all. There is a shallow ridge situated medially in the floor of the ventral canal posteriorly.

ML1190-25
This vertebra is unpublished and it is the smallest in overall dimensions of the sample. The right anterior part of the centrum is damaged, with most of the anterior facet gone too. The neural arch is missing. There is some minor erosion in the posterior facet (Fig S12). The left anterior lateral surface presents a small depression, not morphological but either pathological or taphonomical. Its shape is trapezoidal, with the long sides pointing anterodorsally to posteroventrally. Its length is 13 mm and the width is 9 mm. In the right side of the centrum, slightly more posterior, but still in the anterior part of the centrum there is a small bump of a similar nature, with the long axis located precisely anteroposteriorly. It is 10 mm long and 5 mm wide. Overall both lateral surfaces present a pleurocentral depression.
The ventral surface does not present any ridge or keel on in, just the buttresses of the chevron facets that project anteriorly for about 17 mm.
The posterior facet of the centrum is slightly eroded dorsally and it is concave in shape. The dorsal and lateral rims of the facet are eroded, but it is possible to say that it was nearly square shaped.

ML1190-240
This caudal vertebra was recovered during the excavation of June 2020 and is one of the best preserved in the sample. The centrum is complete and mostly intact, as well the prezygapophyses and the neural arch. Most of the postzygapophyses and the first centimeters of the neural spine are reasonably complete too, despite the damages caused by the gypsum that cuts part of the neural arch ( Fig S12).

ML1190-241
This caudal vertebra was recovered during the excavation of June 2020. The right prezygapophysis is eroded away and the right side of the neural arch is mostly gone, being attached thanks to sediment. The left side (in posterior view) is much better preserved with just minor cracks on it. The postzygapophyses are highly damaged, but still recognizable ( Fig S14).

ML1190-275
This half caudal vertebral centrum was discovered in June of 2020. Given the lack of chevron facets it is interpreted as an anterior part of the centrum. Interestingly it does not fit with ML1190-20, being smaller in size. The anterior facet is mostly square shaped and concave. The floor of the neural canal presents some eroded ridges and pits. During all the preserved length it diminishes its mediolateral width, reaching 4 mm at the point where the centrum is broken. The lateral surfaces are flat and the ventral surface presents the hint of two ridges delimiting a groove, about 7 mm wide. The broken surface of the centrum is hollow, but does nor present a mirror image of the neural canal as in ML1190-20, instead the cavity is oval in shape, with the long axis dorsoventrally oriented. The walls of the vertebra around this cavity are around 7 mm thick (Fig S15). r.n.c, ridges in the floor of the neural canal, n.c, neural canal, geo, geode structure.